Comment 7 for bug 1642098

> As far as I know we've only seen corruption in single volumes, not across multiple volumes.

If you are using incremental backups, doesn't a loss of any volume entail a loss of the multiple volumes afterwards which depend on it...?

In any case, I can safely say that loss of individual volumes, rather than bit errors in networking (how would that even happen given TCP checksums etc?), is the only kind of corruption I've ever had with duplicity, as that is typically how failure happens on my DVDs/BD/HDDs - the medium goes bad and now you cannot read a file, as opposed to reading the file with 1 bitflip. This would be irrecoverable with anything less than 100% redundancy. I just burned a set of BDs, and a number of the split files couldn't be read back; which is fine, as I could reburn them onto the next BD and I had generated a very large amount of PAR2 redundancy over the *entire* set of split files, as PAR2 is intended to be used, and lost files can easily be reconstructed.

There should at least be a warning in the man page about the generated PAR2 being useful only for correcting bitflips or very small amounts of corruption and not major data loss, as the current 'a note on par2 wrapper backend' misleadingly implies that PAR2 is being used normally.